Evaluation of Image Quality in Digital Breast Tomosynthesis

Abstract

The most widely used imaging modality for early detection of breast cancer is mammography. However, it has limitations, since it obtains two-dimensional images of a three-dimensional structure, generating tissue overlap. The implementation of digital breast tomosynthesis (DBT) makes it possible to minimize this condition, allowing for more accurate diagnoses. However, to ensure that images with diagnostic quality are acquired, it is important to perform periodic evaluations on the equipment. In Brazil, there are few studies on quality control in DBT, and specific limits have not yet been established in national standards. Therefore, the tests performed in this study were based on international protocols, such as the Protocol of the Spanish Society of Medical Physics (SEFM), the Protocol for Quality Control of the physical and technical aspects of digital Breast Tomosynthesis systems of the European Reference Organisation for Quality Assured Breast Screening and Diagnostic Services (EUREF) and those used by the National Health Service Breast Screening Programme (NHSBSP), in order to establish a local baseline. Quantitative evaluations were performed by calculating the signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR), and qualitative evaluations were performed by visually inspecting for artifacts and inhomogeneities; and by assessing the system's ability to identify groups of microcalcifications, masses, and fibers using the Phantom ACR (American College of Radiology). The results indicate good agreement in the qualitative and quantitative tests, with homogeneous images throughout the reconstructed volume. A reduction in the CNR was observed in the reconstructed volumes compared to the raw data, possibly attributed to the reconstruction algorithm. The CNR varied satisfactorily, remaining below the 20% limit, demonstrating the equipment's ability to handle different dense areas. In the repeatability and DGM tests, the results were below the reference limits, and the performance in the ACR simulator remained within the manufacturer's specifications.